We prospectively evaluated the accuracy of computerized impedance plethysmography (CIP) in the diagnosis of asymptomatic deep vein thrombosis (DVT) in 246 consecutive high-risk patients scheduled for hip surgery, with bilateral venography used for comparison. The CIP was performed as a surveillance program every third day. If the CIP remained negative, bilateral venography was performed on postoperative day 10 +/- 1 or on day of treatment 14 +/- 1 in nonoperated-on patients. If the CIP became positive, venography was performed within 24 hours. The sensitivity and specificity of CIP for proximal and distal DVT were 19% (confidence interval [CI], 13% to 24%) and 91% (CI, 87% to 94%), respectively. The positive and negative predictive values were 52% (CI, 38% to 65%) and 70% (CI, 65% to 74%), respectively. The sensitivity and specificity of CIP for proximal DVT were 24% (CI, 13% to 34%) and 90% (CI, 87% to 94%), respectively; the positive and negative predictive values were 31% (CI, 20% to 51%) and 87% (CI, 83% to 90%), respectively. We conclude that, because of its low sensitivity, CIP cannot be used in the surveillance of DVT in high-risk patients or for outcome measurements in clinical trials on DVT prophylaxis.